Interlude: Conveyor Belts, Evidence-Based Trust, and Burrito Paragraphs
You might have noticed that I haven’t posted a new YouTube video for a few weeks. I haven’t dropped the ball. But I have hit pause.
A few weeks ago, on a filming day, I had a family emergency (everyone’s okay). Because I create and edit all of the videos myself, I know how many hours go into producing them, and I knew I wouldn’t be able to post one that week. I was disappointed—and I felt like I was letting you down.
But there was silver lining: it forced me off the conveyor belt of constantly “producing.”
And once the conveyor belt stopped, I realized how easily we can fall into the trap of “do more.” I was already writing this weekly newsletter, posting daily on LinkedIn, and creating weekly YouTube videos. On top of that, people were asking me to create YouTube shorts, post carousels and videos for LinkedIn, and even start an Instagram account.
All of this takes time and energy. And as much as I want to support you in every way I can, chasing “more” meant less quality, less creativity, and less value for you. And that’s what really felt like I was letting you down.
So I stopped the conveyor belt for a few weeks to spend some real time thinking about how to bring you the most value. I’m not sure yet whether I’ll turn it back on as is, or build something different. But I do know this: you’ll be the reason for whatever comes next.
Till then, I’ll still be sending this newsletter every week and posting on LinkedIn. And I’ll share what’s next soon.
Now onto this week's round-up...
Round-up
Reading
Medical students are using a popular research tool to pump out misleading studies
“…the combination of inexperienced users and TriNetX’s push-button analysis tools can lead to shoddy publications, which often do not correct for potential biases that can make treatments appear more effective than they are. And because the data can be analyzed so quickly, users can easily cherry-pick positive results for publication, a practice known as p-hacking.”
Controlling AI hallucinations: Building evidence-based trust in clinical and scientific workflows
”In a 2025 cross-industry survey, 44% of organisations reported experiencing negative consequences from generative AI use, with average financial losses of $4.4 million per incident. In the pharmaceutical and healthcare sectors, where precision is paramount and human lives are directly impacted, these consequences manifest most dangerously as AI hallucinations, fabricated or inaccurate outputs presented with unwarranted confidence.”
A call for clarity: a unified checklist for reporting use of large language models in writing scientific manuscripts
”We propose a structured framework to help authors and journal editors and editorial offices distinguish between acceptable and unacceptable uses of generative AI in scientific publications. To operationalize this, we introduce a novel online reporting tool that guides authors in documenting AI use and generates a standardized, citable disclosure statement to ensure transparency and accountability.”
Watching
How to Write Simply
In this Instagram post, Harry Dry shares how a good paragraph is like a burrito, how good sentences don’t talk awkwardly to each other, and why he writes four or five versions of paragraphs while editing.
Quote
“Attention isn't free. It's the most valuable thing you spend.” –Shane Parrish
Thank you so much for reading.
Warmly,
Crystal
